Screening of Postoperative Pulmonary Complications by Electrical Impedance Tomography

• Study cycle:

1. Check of criteria of inclusion/exclusion
2. Informed consent
3. Recording preoperative data via paper case report form
4. Preoperative EIT measurement
5. Postoperative EIT measurement
6. Recording postoperative data via paper case report form Patients therapy and any diagnostic investigation during hospital stay (apart from EIT measurement) are performed appropriate to patients individual clinical needs and are independent from participation in this clinical trial. As PPC may result from surgical complications (e.g. intra- /postoperative bleeding, anastomotic leak, infection of the wound, injury of neighboring organs) they are monitored.

• Statistical analysis:

Sample size:

After consultation of Johannes-Gutenberg-University Institute of Medical Biometry and Epidemiology the following procedure was developed: Pre-analysis with 20 patients, then determination of definitive sample size (advised 80 patients).

Population:

Any registered patient is included in population even if no study-treatment was performed. To fulfill Pre-Protocol-Population-criteria EIT-measurement has to be performed before and after elective epigastric surgery and information about PPC must be revealable from patients medical file/records.

• Analysis of efficacy:

1. Primary objective: (A) Incidence of any PPC (postoperative pulmonary complication) until postoperative day 7; Acquisition of PPC might require diagnostic investigation during hospital stay; investigations are only performed appropriate to patients individual clinical needs. As PPC are defined:

• I: desaturation, increased oxygen demand (data from electronic health record)
• II: dys-/atelectasis, pleural effusion, pneumothorax, pulmonary vein congestion (data from thoracic x-ray)
• III: detected respiratory infection/pneumonia, pulmonary edema, Aspiration, non-invasive/CPAP-Ventilation, necessity of pleural drain/of bronchoscopy (data from electronic health record, x-ray)
• IV: Reintubation (data from electronic health record), ARDS (according to Berlin Definition) (B) regional Distribution of Ventilation (Analysis of EIT raw data pre- and postoperative in 4 regions of interest: ventral, mid-ventral, mid-dorsal, dorsal; correlation between PPC and difference in regional Distribution of Ventilation at defined moment)

1. Secondary objective: post-hoc Analysis (see down Secondary Outcome Measure) Average, Median, Quartile, Standard Deviation are detected and shown in Box-plots (type of graphic). Explorative comparison of parameters is effected by t-Test, Mann-Whitney-U-Test, Fisher-Test dependent on normal distribution.

• Quality assurance and data management Design, execution and analyzation of the study is ICH-GCP-conform (Good-Clinical-Practice-Guideline of the International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use). All responsible persons are obliged to good clinical practice. All changes in study protocol are to be authorized by responsible Ethics Committee. The P.I. is responsible for source data verification to assess the accuracy, completeness, or representativeness of registry data by comparing the data to external data sources (e.g., medical records, paper or electronic case report forms).

Prior to written informed consent all patients are to be informed about character and consequences of the study. With their signature patients permit collection and use of source data. Every patient receives an Identification code for pseudonymization of data. Analyzation and publication of data is effected anonymously only. Data collection is mainly provided in an electronic data base, besides in a paper-based Case Report Form. All electronic and paper-based data and archived in our Clinical Research Center and are accessible for authorized persons (Monitor, Auditor). The study protocol does not provide audits.

Amendment after Pre-analysis:

We conducted a preliminary study of 23 patients for sample size calculation, aligned with the goal of predicting probability of relevant PPC. Power calculation was based on the assumption that 50 % of patients would develop relevant PPC. We performed a power analysis using the Wilcoxon test. The assumption was that AUC of EIT-parameters for distinguishing PPC-affected patients from non-affected patients would be at least 0.65. In order to demonstrate that AUC is indeed above 0.5, a Wilcoxon test can be applied. AUC=0.65 translates into P(X<Y)=0.65. To demonstrate this at the 5%-significance level with a power of 80%, 116 patients are needed. Considering that approximately 10% of patients may not be evaluable, a total sample size of 130 patients was determined. Interim analysis revealed, that incidence of relevant PPC in the study cohort was lower than assumed in sample size calculation. Therefore a higher sample size was necessary than expected. Furthermore, recruiting was slow. We therefore switched from a confirmatory to a descriptive approach.